Reading through the specs these appear to be the same with the following:
a) 4396 can sample up to 192khz whereas 4393 is 96khz. In an application for example trying to mod a dcx with this chip, I believe supports only a max of 96khz anyway?
b) Potential problem is the 4396 output is 2.8v/5.6v whereas 4393 is 2.4v/4.8v
c) 4396 sports a linear 256 step digital attenutation, has anyone interfaced to this?
So the question is why bother modding and how are people claiming improved performance (other than louder due to voltage LOL)
a) 4396 can sample up to 192khz whereas 4393 is 96khz. In an application for example trying to mod a dcx with this chip, I believe supports only a max of 96khz anyway?
b) Potential problem is the 4396 output is 2.8v/5.6v whereas 4393 is 2.4v/4.8v
c) 4396 sports a linear 256 step digital attenutation, has anyone interfaced to this?
So the question is why bother modding and how are people claiming improved performance (other than louder due to voltage LOL)
The AK4396 has an incredibly quiet background.
That what I remember reading somewhere...
Here it is, was from someone at AKM.
AudiogoN Forums: Hi-fi Wi-fi device
Cut&Past:
07-26-06: rkulavik@akm.com
AKM is the worlds largest DAC Supplier, this I would add does not make you the best. However,the part Alex is talking about is the AK4396 this DAC is a large departure from other delta sigma DAC's designed by us and otherslike BB, ADI and Cirrus. The AK4396 is an entirely new modulator, pioneered and patened by AKM that does something unique. In the past many of the old phillips and BB parts were R-2R based products. These older products were looked upon as some of the best. One of the reasons was high freq noise. In older R-2R parts high freq noise was not present. In all delta sigma parts prior to the AK4396 everone fought high freq noise caused by the delta sigma modulator, putting large filters and other things to attempt to solve a problem created by the delta sigma design. The AK4396 today effectively does not have any high freq noise caused by the modulator, it is over 60dB better than the nearest Cirrus and BB devices. All this high freq noise can do many things down stream and end up as audible artifacts. That is the Miracle we belive is making the difference today. This part gives you the performance and linearity of a delta Sigma part while giving you the noise of an R-2R part. Something that was never previously available.
I recommend taking a listen to the products using the AK4396 such as APL's and the Slim Devices box. I think you will be plesently surprised. The AKM spends a lot of time and money working on sound quality it is something that makes working for AKM very enjoyable but it is also demanding. We listen to everything we produce and solicite outside very often for impressions on sound quality. However at the end of the day your own ears have to make the call.
Regards,
Richard Kulavik
AKM Semiconductor
rkulavik@akm.com
That what I remember reading somewhere...
Here it is, was from someone at AKM.
AudiogoN Forums: Hi-fi Wi-fi device
Cut&Past:
07-26-06: rkulavik@akm.com
AKM is the worlds largest DAC Supplier, this I would add does not make you the best. However,the part Alex is talking about is the AK4396 this DAC is a large departure from other delta sigma DAC's designed by us and otherslike BB, ADI and Cirrus. The AK4396 is an entirely new modulator, pioneered and patened by AKM that does something unique. In the past many of the old phillips and BB parts were R-2R based products. These older products were looked upon as some of the best. One of the reasons was high freq noise. In older R-2R parts high freq noise was not present. In all delta sigma parts prior to the AK4396 everone fought high freq noise caused by the delta sigma modulator, putting large filters and other things to attempt to solve a problem created by the delta sigma design. The AK4396 today effectively does not have any high freq noise caused by the modulator, it is over 60dB better than the nearest Cirrus and BB devices. All this high freq noise can do many things down stream and end up as audible artifacts. That is the Miracle we belive is making the difference today. This part gives you the performance and linearity of a delta Sigma part while giving you the noise of an R-2R part. Something that was never previously available.
I recommend taking a listen to the products using the AK4396 such as APL's and the Slim Devices box. I think you will be plesently surprised. The AKM spends a lot of time and money working on sound quality it is something that makes working for AKM very enjoyable but it is also demanding. We listen to everything we produce and solicite outside very often for impressions on sound quality. However at the end of the day your own ears have to make the call.
Regards,
Richard Kulavik
AKM Semiconductor
rkulavik@akm.com
Last edited:
I don't buy it, neither does the datasheet, stating:
"The internal switched-capacitor filters attenuate the noise generated by the delta-sigma modulator beyond the audio passband"
Any why would the filter be exactly the same, as the presumably so much different AK4393? That doesn't make sense... And for having such a revolutionary new modulator, AKM is fairly quiet about it, not even mentioning it..
Why is AKM about the only one that is not specing out of band noise, not even an simple picture...?
4real, well that's what is stated by the guy from AKM in that post I cut & pasted from the other site, not myself.
Though I did swap the AK4393 out of a SRC2496 I had for the AK4396 and the AK4396 had such a black background I was stunned by it, unfortunately I killed the chip (my own fault) before I could optimise the output section of the SRC2496.
As for those questions, you would need to ask AKM themselves
Though I did swap the AK4393 out of a SRC2496 I had for the AK4396 and the AK4396 had such a black background I was stunned by it, unfortunately I killed the chip (my own fault) before I could optimise the output section of the SRC2496.
As for those questions, you would need to ask AKM themselves
Why Bother with a AK4396
The 4396 data sheet suggests that it's differential SCF output eliminates the need for a coupling capacitor, Is this not a good enough reason for changing to the 4396?
Also, a major "Gotcha" with upgrading a DCX or DEQ to a AK4396 has to be the higher output voltage - this chip is not a drop in replacement - one will have to change the driver stage to compensate for the higher output.
The 4396 data sheet suggests that it's differential SCF output eliminates the need for a coupling capacitor, Is this not a good enough reason for changing to the 4396?
Also, a major "Gotcha" with upgrading a DCX or DEQ to a AK4396 has to be the higher output voltage - this chip is not a drop in replacement - one will have to change the driver stage to compensate for the higher output.
That's why I was thinking one possible reason why people think it is improved sound (although I respect the other claims numerous ppl have made and will still do this mod myself.) There is evidence that people in listening tests think that louder is better sometimes in an A/B test where B has a slightly higher output.
I didn't say quieter just a perception of better. I think others will agree that a small increase in volume can sometimes trick a person into thinking better. In your case you are more of an audiophile I'm sure and can tell the difference!
Anyway lets stick to the outstanding question... does the 4393 not have the same "differential SCF output" even though stated in different words in the spec sheet?
I spent last Sunday removing the AK4393 and installed a AK4396 in my DEQ. I haven't had much time to do any serious comparisons but the higher output level did throw me for a loop. I re-ran the room correction function and dropped the DEQ's level by 3db but I am still experiencing some output stage clipping (especially on some Grammy Award winning - compressed to hell recordings!)
This is my first attempt at ripping out a SMD chip so I did several practice runs on some scrap pc boards. I initially tried using a needle from a sewing machine (I heated and bent the needle to a 45 degree angle) This worked well on the memory chips I practiced removing. I pushed the needle under the chip legs and heated the legs (and needle) with a soldering iron. The legs jumped clear of the solder as the needle passed on to the next set of legs. I removed about 6 chips in total and did not damage any PC tracks. Once the chips were out, I used a length of solder wick to remove the excess solder and then cleaned the board with Isopropyl Alcohol.
At this point, I felt confident enough to attack the DEQ. I gave up on my "needle technique" after lifting a few legs as I could not manipulate the needle without touching the adjacent chips. I was now at the point of no return and switched to plan B which was to use a Stanley Knife to gently cut through the 9343's legs. This took a great deal more time than the "needle technique" but did not make contact with any of the adjacent chips.
Once the first side of the chip was free, I broke off the remaining legs with a to and fro motion. Removing the residual solder and chip legs proved to be very easy with the solder wick. I proceeded to clean the board with 99.953% Isopropyl Alcohol and then placed the AK9396 on the board.
It was pretty much impossible to locate the chip on the solder pads with 100% accuracy - I will resort to using a SMD glue the next time I perform this operation. When the chip was reasonably positioned, I proceeded to add a minute drop of solder to the first leg and this turned out to be a huge mistake! My minuscule dab of solder turned out to be way too much and bridged across the adjacent legs of the 4396. I had to remove this excess solder using solder wick once again. In hindsight, there is sufficient residual solder on the pads to make a decent joint and it is not necessary to add any solder when soldering in the new chip.
After cleaning the board, I had to be sure that there were no solder bridges or cold joints. My 10x eye loupes combined with my poor eye-sight were not up to the QC task so I resorted to taking a number of pictures with my Lumix Camera and did my final inspection on my 24" monitor with the pics on full zoom. As it happens, there was just one suspect joint which I was able to fix with little trouble.
The DEQ powered up with no issues and appeared non-the-wiser, showing no signs of silicon organ rejection and I was able to breathe again...
I left the DEQ powered on since Sunday and will begin my listening sessions later in the week when I have some time.
I'll post my impressions sometime soon!
Cheers
Jaimo
This is my first attempt at ripping out a SMD chip so I did several practice runs on some scrap pc boards. I initially tried using a needle from a sewing machine (I heated and bent the needle to a 45 degree angle) This worked well on the memory chips I practiced removing. I pushed the needle under the chip legs and heated the legs (and needle) with a soldering iron. The legs jumped clear of the solder as the needle passed on to the next set of legs. I removed about 6 chips in total and did not damage any PC tracks. Once the chips were out, I used a length of solder wick to remove the excess solder and then cleaned the board with Isopropyl Alcohol.
At this point, I felt confident enough to attack the DEQ. I gave up on my "needle technique" after lifting a few legs as I could not manipulate the needle without touching the adjacent chips. I was now at the point of no return and switched to plan B which was to use a Stanley Knife to gently cut through the 9343's legs. This took a great deal more time than the "needle technique" but did not make contact with any of the adjacent chips.
Once the first side of the chip was free, I broke off the remaining legs with a to and fro motion. Removing the residual solder and chip legs proved to be very easy with the solder wick. I proceeded to clean the board with 99.953% Isopropyl Alcohol and then placed the AK9396 on the board.
It was pretty much impossible to locate the chip on the solder pads with 100% accuracy - I will resort to using a SMD glue the next time I perform this operation. When the chip was reasonably positioned, I proceeded to add a minute drop of solder to the first leg and this turned out to be a huge mistake! My minuscule dab of solder turned out to be way too much and bridged across the adjacent legs of the 4396. I had to remove this excess solder using solder wick once again. In hindsight, there is sufficient residual solder on the pads to make a decent joint and it is not necessary to add any solder when soldering in the new chip.
After cleaning the board, I had to be sure that there were no solder bridges or cold joints. My 10x eye loupes combined with my poor eye-sight were not up to the QC task so I resorted to taking a number of pictures with my Lumix Camera and did my final inspection on my 24" monitor with the pics on full zoom. As it happens, there was just one suspect joint which I was able to fix with little trouble.
The DEQ powered up with no issues and appeared non-the-wiser, showing no signs of silicon organ rejection and I was able to breathe again...
I left the DEQ powered on since Sunday and will begin my listening sessions later in the week when I have some time.
I'll post my impressions sometime soon!
Cheers
Jaimo
Cool Jaimo! I should be posting my experience soon; I'm in a very similar boat. I've not done any SMD rework either. I am a super impatient type and will have to force myself to practice first. At first I was going to go the hot air route and buy a cheap station. Then I saw you can get SOP 5.6mm X ?mm nozzles and thought that would work real good but then $ started to add up.
Instead I got ahold of something called chipquick and I think it will do a great job fast on removal. For soldering back on I've read a lot of info suggesting that super fine tip isn't necessarily the best. I'm going to try the "drag" technique using a an angled tip, making sure everything is spot clean and well fluxed with noclean flux and fine solder and try soldering a whole side of pins in one 1-second stroke.
Instead I got ahold of something called chipquick and I think it will do a great job fast on removal. For soldering back on I've read a lot of info suggesting that super fine tip isn't necessarily the best. I'm going to try the "drag" technique using a an angled tip, making sure everything is spot clean and well fluxed with noclean flux and fine solder and try soldering a whole side of pins in one 1-second stroke.
Fair enough, my misunderstanding
Myself I would only do it again if I intended using the chip with a transformer direct output.
I'm very interested in opinions of those giving this chip a go
Hot air station is the best way to do it. Hot air gun used for shrink tubbing, with temp. control, could be a cheep way of doing it. Try to make some smaller nozzle and that should do it. Your heat gun really needs fine temp. control and only the best heat guns have that and are capable of going low enough for this purpose. Next, for soldering chip back in place, you do not need a glue. Use gel flux, that will hold chip in place and help you with soldering.
Do not worry if your solder spills on several legs, just use the finest solder wick and remove the excess solder. Sometimes the thinnest soldering tips are not good because they do not transfer enough heat, so use medium sized tip.
Hope this helps.
Do not worry if your solder spills on several legs, just use the finest solder wick and remove the excess solder. Sometimes the thinnest soldering tips are not good because they do not transfer enough heat, so use medium sized tip.
Hope this helps.
4396 vs 4393 is a clear and noticeable improvement, particularly in transparency department. The improvement is not day - night, much more subtle than that, but still noticeable. The sound is more sophisticated and natural in high Fqs. That difference you will never be able to find in specs, but you will be able to hear it.
I am not familiar with Jan's active output boards. This is a first time that I hear about differences in output. In my case that is OK, because I am using transformer coupled output with symmetrical jFet buffers. Quite honestly I didn't noticed any difference in that department. The only difference I found was the one described above, and that is something that is not influenced by output level, but by the character of the output signal. I do remember though, that Jan has made that conversion from 4393 to 4396 chip and I do not think that he expressed any concerns about output level with his boards. It was interesting that his and some other people's findings that made the conversion, were very similar to what I was experiencing and reporting.
In addition, manufacturer states that these two DAC chips are straight replacement for each other.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.